semocean.com
如何使用机器学习解决实际问题-以关键词相关性模型为例 | Dustinsea
http://semocean.com/如何使用机器学习解决实际问题-以关键词相关性模
这是一个典型的搜索问题，具体的从输入query，到触发，到排序等会涉及到很多因素，例如如何查倒排，如果处理地域因素等 要提升搜索的质量，我们首先需要保证输入的query和推荐出来的推荐词的相关性，此处我们要解决的主要问题， 就是如何快速，准确地判断两个关键词 输入query和推荐词 的相关性，需要特别注明的是，我们主要的 目标是让用户觉得该产品结果很靠谱， 所以该处我们仅考虑字面相关性，更多的语意扩展该模型并未考虑. 相关性也是所有搜索问题的基石，只不过在不同的系统中使用方式不一样， 在一般的搜索中，相关性占有较大权重， 排序基本就以相关性为依据 在商业系统中，相关性则经常作为搜索展现的门槛用于控制商业推广结果的质量 如果仅考虑CTR， 用户搜索鲜花快递时，给用户展现艳照门的结果，CTR会更高，但相关性较差。 本文以下内容会从数据准备， 特征选择， 模型选择， 模. 人工标注 优点是质量较高，噪音较少 缺点是标注结果和标注者本身的认识相关，例如在搜索引擎中，判定苹果和手机的相关性，对于年轻人，一般都认为相关 但对于比较多的老人，可能认为不相关 另外一个缺点就是人工获取标注的成本较高. 现在特别喜欢使用树...
lorehound.com
QuakeCon 2015: Case-Mod Sampling – All your MMO & MOBA Needs – Lore Hound
http://lorehound.com/news/quakecon-2015-case-mod-sampling
All your MMO and MOBA Needs - Lore Hound. MMO Game and MOBA Opinions, Videos and Reviews. Advertise with Lore Hound. QuakeCon 2015: Case-Mod Sampling. July 24, 2015. Conventions are cool. So are huge LANs. What can’t you miss at the 20th anniversary QuakeCon? Well- One reason: The dedicated few who spend months to years on their case modifications. We’ve spent some of our own time gathering them up so you don’t have to bother. And stay tuned until the end! Captured as best as I could manage, this mod doe...
pancakesforrecess.com
Sunday.104 | Pancakes for Recess
http://pancakesforrecess.com/2014/09/07/sunday-104
September 7, 2014. Two years ago: The Beginning. Three years ago: The Kissing Hand. Four years ago: Goodbye and Hello. One Response to Sunday.104. August 10, 2015 at 1:00 AM. Leave a Reply Cancel reply. Your email address will not be published. Required fields are marked *. You may use these. A href= title= abbr title= acronym title= b blockquote cite= cite code del datetime= em i q cite= strike strong. See my 1st grade pic. No public Twitter messages. A Cup of Jo. The O'Brien House. You Are My Fave.
